The inventors of the present invention have recognized the success of semiconductor devices has been mainly driven by an intensive transistor down-scaling process. However, as field effect transistors (FETs) approach sizes less than 100 nm, physical problems such as short channel effect begin to hinder proper device operation. For transistor based memories, such as those commonly known as Flash memories, other performance degradations or problems may occur as device sizes shrink. With Flash memories, a high voltage is usually required for programming of such memories. However, as device sizes shrink, the high programming voltage can result in dielectric breakdown and other problems. Similar problems can occur with other types of non-volatile memory devices other than Flash memories.
The inventors of the present invention recognize that many other types of non-volatile random access memory (RAM) devices have been explored as next generation memory devices, such as: ferroelectric RAM (Fe RAM); magneto-resistive RAM (MRAM); organic RAM (ORAM); phase change RAM (PCRAM); and others.
A common drawback with these memory devices include that they often require new materials that are incompatible with typical CMOS manufacturing. As an example of this, Organic RAM or ORAM requires organic chemicals that are currently incompatible with large volume silicon-based fabrication techniques and foundries. As another example of this, Fe-RAM and MRAM devices typically require materials using a high temperature anneal step, and thus such devices cannot be normally be incorporated with large volume silicon-based fabrication techniques.
Additional drawbacks with these devices include that such memory cells often lack one or more key attributes required of non-volatile memories. As an example of this, Fe-RAM and MRAM devices typically have fast switching (e.g. “0” to “1”) characteristics and good programming endurance. However, such memory cells are difficult to scale to small sizes. In another example of this, for ORAM devices reliability of such memories is often poor. As yet another example of this, switching of PCRAM devices typically include Joules heating and undesirably require high power consumption.
From the above, a new semiconductor device structure and integration is desirable.